[jira] [Created] (ARROW-13676) [C++] Coredump writing Arrow table to Parquet file

["Shuai Zhang (Jira)" <ji...@apache.org>]

Shuai Zhang created ARROW-13676:
-----------------------------------

             Summary: [C++] Coredump writing Arrow table to Parquet file
                 Key: ARROW-13676
                 URL: https://issues.apache.org/jira/browse/ARROW-13676
             Project: Apache Arrow
          Issue Type: Improvement
          Components: C++, Parquet
    Affects Versions: 4.0.1, 5.0.0, 6.0.0
            Reporter: Shuai Zhang
         Attachments: callstack.txt

I'm suffering a random appeared coredump issue converting user data from Gogole Protobuf format to Apache Parquet file via Apache Arrow C++ project. The problem could be stable reproduced with ASAN check enabled for specified user data. The callstack from ASAN check is exactly same as the coredump callstack (posted in attachment file, compiled with apache-arrow-4.0.1 built without jemalloc). 

I made some initial investigations:

# The direct constructed Arrow table would trigger this issue. Clone it in different way would yield different result, despite all of them are equal via `table.Equals(other)` method.
## Serialize then deserialize the table was safe.
## CombineChunks didn't help.
## Clone with TableBatchReader didn't help.
## CombineChunks or TableBatchReader cloning on deserialized table was still safe.
# Different environment would trigger this problem, I think the issue is not related to glibc
## Debian 8 + gcc 4.9.2
## Debian 9 + gcc 6.3.0
## Debian 11 + gcc 10.2.1
## Ubuntu 20.04 LTS + clang 12.0.1

I still need to cleanup security policy sensitive codes & data to provide a way to stable reproduce this issue. I'll come back a few days later.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)